On 5 August 2015 05:53:44 CEST, Simon Glass sjg@chromium.org wrote:
I think this commit message might need a few tweaks. This is not a regression as mentioned elsewhere. Also I don't thing it means CONFIG_DM here. Maybe CONFIG_DM_USB?
Yes you are right. It's about DM_USB of course, sorry.
There was a report that this does in fact not work ("CONFIG_DM_ETH USB_ETHER_ASIX Reception Issue on Tegra"). does this patch fix that problem? I'm not sure how but the above suggests that it does.
No, those are two entirely different issues. Fact is right now almost nobody defines DM_ETH but e.g. on Tegra DM_USB is set which currently renders on-module Ethernet defunct on both Colibri T20 as well as Colibri T30. This patch fixes just that and Ethernet will work again.
The logic here was really ugly (sorry!). However even so I'm struggling to understand what this patch actually changes. It seems that the #error is activated:
#if !CONFIG_DM_USB && CONFIG_USB_HOST_ETHER && CONFIG_DM_ETH
which looks the same as before.
No, it's not only about the error message. You missed the usb_host_eth_scan() call! Without that being called in the old non-DM_ETH case it simply won't work! Which is the case in current mainline U-Boot on Colibri T20/T30.
So overall I'm a bit confused. Hoping we can figure this out soon. The patches to enable driver model USB Ethernet and Asix were applied a few weeks ago and there may well still be problems. As I mentioned I did not test this on Tegra hardware unfortuantely.
As mentioned before. That's a different story but as DM_ETH is not currently enabled on any Tegras I consider this a secondary issue